[uf-discuss] Non-HTTP tags

[Brian Suda]

On 07/01/2008, Tom Morris <bbtommorris at gmail.com> wrote:
> The rel-tag specification says that tags ought to be HTTP URIs.
>
> In a page I'm authoring at the moment, I'm using non-HTTP URIs as tags

--- this is interesting. I can´t speak for the original intent of only
HTTP, but one reason might be transparency. Anyone who wants to know
what the tag "Turkey" might mean, can dereference the URL and get a
page that further explains that term. This is why both the tag and
tag-space are important.

If we are using non-HTTP links, then some of that transparency is
lost. irc://irc.freenode.net/microformats could be deferenece
(debatable if the user is not familiar or don´t have an IRC client)
then once you are there, there wouldn´t be much in the way of
human-readable text to further explain what "microformats" means.
Again, debatable because the channel would be full of people who could
explain what that term means.

> ... there is some kind of philosophical
> justification for the current HTTP only policy for tags. It seems to
> me that so long as the target resource URI conforms to the tag URI
> structure, it shouldn't matter what protocol it uses.

--- i would say that it would need to dereference to something that
explains the term more, not all protocols would do this.

> Also, any thoughts on a rel-tag test suite? Specifications say,
> implementations show, tests prove, remember. :)

--- there is a basic test suite at hg.microformats.org, but the only
rel-tag tests are in conjunction with compound microformats. A rel-tag
test suite would be welcomed.

-brian

-- 
brian suda
http://suda.co.uk